v1.8.2

• GH-1571: Remove trimming inside individual chunks.

  Trimming Chunks (always from the left) causes a lot of internal work with only limited benefit since we manage visibility with stream::Views on top of Chunks anyway.

  This patch removes trimming inside Chunks so now any trimming only removes Chunks from Chains, but does not internally change individual Chunks anymore. This might lead to slightly increased memory use, but callers usually have that data in memory anyway.

• GH-1549: GH-1554: Fix potential infinite loop when trimming data before stream.

  Previously we would trigger an infinite loop if one tried to trim before the head chunk of a stream. In praxis this seem to have been no issue due to #1549 and us emitting way less calls to trim than possible.

  This patch adds an explicit check whether we need to trim anything, and exits the low-level function early for such cases.

• GH-1550: Replace recursive deletion with explicit loop to avoid stack overflow.

• GH-1549: Add feature guards to accesses of a unit's __position.

  Access of __position triggers a random access functionality. In order to distinguish our internal uses from accesses due to user code, most access in our generated code should be guarded with a feature constant (if or ternary).

  In this patch add proper guards for a couple instances where we did not do that correctly. That mishap caused all units with containers to be random access (even the root unit) which in turn could have lead to e.g., unbounded memory growth, or runtime overhead due to generation and execution of unneeded code, or expensive cleanup on very large untrimmed inputs.

• Artificially limit the number of open files.

  This works around a silent failure in reproc where it would refuse to run on systems which huge rlimits for the number of open files. We have seen this hit on huge production boxes.

• GH-1478: Add regression test for #1478.

• Add begin to parser state.

  This patch adds the current begin position to the parser state, and makes the corresponding changes to generated parser functions so it is passed down.

  We already modelled the semantic beginning of the input in the unit, but had no reliable way to keep this up-to-date across non-unit contexts like &parse-from. This would then for certain setups lead to generated code where input and position would point to different inputs which in turn caused offset (modelled as position - input) to be incorrect.

• Expand validator error message.

• Disable a few newer clang-tidy categories.

  The options disabled here and triggered in newer versions of clang-tidy.

• Drop -noall_load linker option.

  We added this linker option on macos. This option was already obsolete, e.g., in the ld manpage:

  -noall_load
          This is the default.  This option is obsolete.
  

  Newer versions of xcode do not know this option anymore and instead generate a hard error.

• Declare Spicy pygments extension as parallel-safe.

  We previously would not declare that the Spicy pygments highlighter is safe to execute in parallel (reading or writing of sources). Sphinx then assumed that the extension was not safe to run in parallel and instead ran jobs sequentially.

  This patch declares the extension as able to execute in parallel. Since the extension does not manage any external state this is safe.

• Use find_package(Python) with version.

  Zeek's configure sets Python_EXECUTABLE has hint, but Spicy is using find_package(Python3) and would only use Python3_EXECUTABLE as hint. This results in Spicy finding a different (the default) Python executable when configuring Zeek with --with-python=/opt/custom/bin/python3.

  Switch Spicy over to use find_package(Python) and add the minimum version so it knows to look for Python3.

v1.8.1

• GH-1475: Incorrect offset computation when using lists and &parse-from
• GH-1464: Add special handling for potential advance failure in trial mode

v1.8.0

New Functionality

Add new skip keyword to let unit items efficiently skip over uninteresting data.

For cases where your parser just needs to skip over some data, without needing access to its content, Spicy provides a skip keyword to prefix corresponding fields with:

module Test;

public type Foo = unit {
  x: int8;
   : skip bytes &size=5;
  y: int8; on %done { print self; }

};

skip works for all kinds of fields but is particularly efficient with bytes fields, for which it will generate optimized code avoiding the overhead of storing any data.

skip fields may have conditions and hooks attached, like any other fields. However, they do not support $$ in expressions and hooks.

For readability, a skip field may be named (e.g., padding: skip bytes &size=3;), but even with a name, its value cannot be accessed.

skip fields extend support for void with attributes fields which are now deprecated.

Add runtime profiling infrastructure.

This add an option --enable-profiling to the HILTI and Spicy compilers. Use of the option does two things: (1) it sets a flag enabling inserting additional profiling instrumentation into generated C++ code, and (2) it enables using instrumentation for recording profiling information during execution of the compiled code, including dumping out a profiling report at the end. The profiling information collected includes time spent in HILTI functions as well as for parsing Spicy units and unit fields.

Changed Functionality

Optimizations for improved runtime performance.

This release contains a number of changes to improve the runtime performance of generated parsers. This includes tweaks for generating more performant code for parsers, low-level optimizations of types in to runtime support library as well as fine-tuning of parser execution at runtime.

---

• Do not force locale on users of libhilti.
• Avoid expensive checked iterator for internal Bytes iteration.
• GH-1089: Allow to use offset() without enabling full random-access support.
• GH-1394: Fix C++ normalization of generated enum values.
• Disallow using $$ with anonymous containers.

Bug fixes

• GH-1386: Prevent internal error when passed invalid context.
• Fix potential use-after-move bug.
• GH-1390: Initialize Bytes internal control block for all constructors.
• GH-1396: Fix regex performance regression introduced by constant folding.
• GH-1399: Guard access to unit _filters member with feature flag.
• GH-1421: Store numerical offset in units instead of iterator for position.
• GH-1436: Make sure Bytes::sub only throws HILTI exceptions.
• GH-1447: Do not forcibly make strong_ref in function parameters immutable.
• GH-1452: Allow resolving of unit parameters before self is fully resolved.
• Make sure Spicy runtime config is initialized after spicy::rt::init.
• Adjustments for building with GCC-13.

Documentation

• Document how to check whether an optional value is set.
• Preserve indention when extracting comments in doc generation.
• Fix docs for long-form of -x flag to spicyc.

v1.5.4

• GH-1436: Make sure Bytes::sub only throws HILTI exceptions.
• Allow building with gcc-13.
• Allow optimizer to remove unused filter functionality in units.
• Avoid expensive checked iterator for internal Bytes iteration.
• GH-1390: Initialize Bytes internal control block for all constructors.
• Do not force locale on users of libhilti.
• Fix potential use-after-move bug.
• GH-1310: Fix ASAN false positive with GCC.
• Skip clang-specific ASAN flags with other compilers.
• Don't instantiate a debug logger if we aren't going to debug log.
• Simplify extract methods.
• Shortcut some offset computations.
• GH-1345: Apply alternative fix for #1345.
• Make printParserState cheaper to call if debug logging is disabled.
• GH-1367: Use unique filename for all object files generated during JIT.
• Fix code generation for -X flow or -X trace.
• Remove potential race during JIT when using HILTI_CXX_COMPILER_LAUNCHER.

v1.7.0

New Functionality

• Support Zeek-style documentation strings in Spicy source code.
• Provide ability for host applications to initiate runtime's module-pre-init phase manually.
• Add DPD-style spicy::accept_input() and spicy::decline_input().
• Add driver option to output full set of generated C++ files.
• GH-1123: Support arbitrary expression as argument to type constructors, such as interval(...).

Changed Functionality

• Search HILTI_CXX_INCLUDE_DIRS paths before default include paths.
• Search user module paths before system paths.
• Streamline runtime exception hierarchy.
• Fix bug in cast from real to interval.
• GH-1326: Generate proper runtime types for enums.
• GH-1330: Reject uses of imported module IDs as expression.

Bug fixes

• GH-1310: Fix ASAN false positive with GCC.
• GH-1345: Improve runtime performance of stream iteration.
• GH-1367: Use unique filename for all object files generated during JIT.
• Remove potential race during JIT when using HILTI_CXX_COMPILER_LAUNCHER.
• GH-1349: Fix incremental regexp matching for potentially empty results.

v1.6.1

Bug fixes

• GH-1356: Add stringification to UnitContext.
• GH-1357: Remove potential race during JIT when using HILTI_CXX_COMPILER_LAUNCHER.
• GH-1349: Fix incremental regexp matching for potentially empty results.
• Fix test Backtrace.comparison for aggressive constant folding.
• Install full nlohmann JSON header as well.
• GH-1345: Fix pathological performance for _haveEod on highly chunked streams.
• Remove performance pessimization introduced in 84fb4f21f494.

v1.5.3

Bug fixes

• GH-1349: Fix incremental regexp matching for potentially empty results.
• Install full nlohmann JSON header as well.
• GH-1345: Fix pathological performance for _haveEod on highly chunked streams.
• cmake/FindGoldLinker: Do not use gold by default
• cmake/FindGoldLinker: Put -fuse-ld=gold before existing flags
• GH-1303: Fix incorrect offset computation in advanceToNextData.

v1.6.0

New Functionality

• GH-1249: Allow combining &eod with &until or &until-including.

• GH-1251: When decoding bytes into a string using a given character set, allow caller to control error handling.

  All methods taking a charset parameters now take an additional enum selecting 1 of 3 possible error handling strategies in case a character can't be decoded/represented: STRICT throws an error, IGNORE skips the problematic character and proceeds with the next, and REPLACE replaces the problematic character with a safe substitute. REPLACE is the default everywhere now, so that by default no errors are triggered.

  This comes with an additional functional change for the ASCII encoding: we now consistently sanitize characters that ASCII can't represent when in REPLACE/IGNORE modes (and, hence, by default), and trigger errors in STRICT mode. Previously, we'd sometimes let them through, and never triggered any errors. This also fixes a bug with the ASCII encoding sometimes turning a non-printable character into multiple repeated substitutes.

• GH-1294: Add library function to parse an address from string or bytes.

• HLTO files now perform a version check when loaded.

  We previously would potentially allow building a HLTO file against one version of the Spicy runtime, and then load it with a different version. If exposed symbols matched loading might have succeeded, but could still have lead to sublte bugs at runtime.

  We now embed a runtime version string in HLTO files and reject loading HLTO files into a different runtime version. We require an exact version match.

• New pack and unpack operators.

  These provide low-level primitives for transforming a value into, or out of, a binary representations, see the docs for details.

Changed Functionality

• GH-1236: Add support for adding link dependencies via --cxx-link.

• GH-1285: C++ identifiers referenced in &cxxname are now automatically interpreted to be in the global namespace.

• Synchronization-related debug messages are now logged to the spicy-verbose stream. We added logging of successful synchronization.

• Downgrade required Flex version. We previously required at least flex-2.6.0; we can now build against flex-2.5.37.

• Improve C++ caching during JIT.

  We improved caching behavior via HILTI_CXX_COMPILER_LAUNCHER if the configuration of spicyc was changed without changing the C++ file produced during JIT.

• hilti::rt::isDebugVersion has been removed.

• The -O | --optimize flag has been removed from command line tools.

  This was already a no-op without observable side-effects.

• GH-1311: Reject use of context() unit method if unit does not declare a context with %context.

• GH-1319: Unsupported unit variable attributes are now rejected.

• GH-1299: Add validator for bitfield field ranges.

• We now reject uses of self as an ID.

• GH-1233: Reject key types for maps that can't be sorted.

• Fix validator for field &default expression types for constness.

  When checking types of field &default expressions we previously would also consider their constness. This breaks e.g., cases where the used expression is not a LHS like the field the &default is defined for,

  type X = unit {
      var x: bytes = b"" + a;
  };

  We now do not consider constness in the type check anymore. Since fields are never const this allows us to set a &default with constant expressions as well.

Bug fixes

• GH-1231: Add special handling for potential advance failure in trial mode.
• GH-1115, GH-1196: Explicitly type temporary value used by &max_size logic.
• GH-1143, GH-1220: Add coercion on assignment for optionals that only differ in constness of their inner types.
• GH-1230: Add coercion to default argument of map::get.
• GH-1234, GH-1238: Fix assertions with anonymous struct constructor.
• GH-1248: Fix stop for unbounded loop.
• GH-1250: Fix internal errors when seeing unsupported character classes in regular expression.
• GH-1170: Fix contexts not allowing being passed inout.
• GH-1266: Fix wrong type for Spicy-side self expression.
• GH-1261: Fix inability to access unit fields through self in &convert expressions.
• GH-1267: Install only needed headers from bundled SafeInt library.
• GH-1227: Fix code generation when a module's file could be imported through different means.
• GH-1273: Remove bundled code licensed under CPOL license.
• GH-1303: Fix potentially late synchronization when jumping over gaps during synchronization.
• Do not force gold linker with user-provided linker flags or when built as a CMake subproject.
• Improve efficiency of startsWith for long inputs.

Documentation

• The documentation now reflects Zeek package manager Spicy feature templates.
• The documentation for bitfields was clarified.
• Documentation for casts from integers to boolean was added.
• We added documentation for how to expose custom C++ code in Spicy.
• Update doc link to commits mailing list.
• Clarify that %context can only be used in top-level units.
• Clarify that &until consumes the delimiter.
• GH-1240: Clarify docs on SPICY_VERSION.
• Add FAQ item on source locations.
• Add example for use of ?..

v1.5.2

Bug fixes

• GH-1291: Quote CMake string to ensure correct argument count. (Nic Boet)
• GH-1273: Replace Lambda class.
• GH-1289: Downgrade required Flex version.
• Fix linker warning on Darwin.
• Bump justrx.

v1.5.1

Bug fixes

• GH-1248: Fix stop for unbounded loop.
• GH-1230: Add coercion to default argument of map::get.
• GH-1143, GH-1220: Add coercion on assignment for optionals that only differ in constness of their inner types.
• GH-1196: Fix validator for field &default expression types for constness.
• GH-1227: Fix code generation when a module's file could be imported through different means.